generate_random_collections

#!/bin/bash -e

if [ -z "$1" ]; then
  echo "Usage:"
  echo "  generate_random_collections <number of collections>"
  echo "Removes all bucket contents, downloads 6 pictures for each album from unsplash.com, creates 6 albums for each collection, giving each collection and each album a random name, uploads them to the original bucket."
  exit 1
else
  webBucket="$(sed -n 's/  "webBucket=\(.*\)"\,/\1/p' dist/config.json)"
  sourceBucket="$(sed -n 's/  "sourceBucket=\(.*\)"\,/\1/p' dist/config.json)"
  resizedBucket="$(sed -n 's/  "resizedBucket=\(.*\)"\,/\1/p' dist/config.json)"
  numberOfCollections=$1
fi

echo "Removing all files in web, source & resized buckets"
for bucket in "$webBucket" "$sourceBucket" "$resizedBucket"; do
  aws s3 rm --recursive "s3://$bucket" >> /dev/null 2>&1 || true
done

prevMD5=()
for (( collectionIndex = 0; collectionIndex < numberOfCollections; collectionIndex++ )); do
  collectionNumber=$RANDOM
  let "collectionNumber %= 1000"
  for (( albumIndex = 0; albumIndex < 6; albumIndex++ )); do
    albumNumber=$RANDOM
    let "albumNumber %= 1000"
    for (( picIndex = 0; picIndex < 6; picIndex++ )); do
      echo "Downloading pic $picIndex for collection $collectionIndex -> album $albumIndex"
      pic="$(mktemp).jpg"
      curl -sL https://source.unsplash.com/random -o "$pic"
      while [[ " ${prevMD5[@]} " =~ " $(md5sum "$pic" | awk '{print $1}') " ]]; do
        curl -sL https://source.unsplash.com/random -o "$pic"
      done
      prevMD5+=("$(md5sum "$pic" | awk '{print $1}')")
      aws s3 cp "$pic" "s3://$sourceBucket/pics/original/Collection $collectionNumber/Album $albumNumber/0$picIndex.jpg" --sse >> /dev/null 2>&1
    done
  done
done